Please see my updated post above.
Also, looking at your particular application of this, I can't quite understand why do you need to worry much about blessed entities (structures). No matter what 'class' name your hash structure has been branded with, in the end you still deal with the same hash structure. The $self variable is simply referencing that 'named' structure. And when you do $self->method(), Perl simply interprets it as FOO::method($self) on the basis that '$self' has been 'labeled' as belonging to the package or being of type 'FOO'. That's all. Nontheless, I can still do this:
package FOO;
sub new {
bless {
key1 => 1,
key2 => 2
};
}
package main;
my $foo = new FOO();
print "Look, first key = '$foo->{key1}'!\n";
Output:
Look, first key = '1'!
Clearly, that $foo object is nothing but a reference to a hash. Simple as that.
_____________________
# Under Construction
| [reply]
[d/l]
[select] |
That works for normal perl, but not for templates in Template Toolkit (TT2), which is what the attempted unblessed object is for use with. TT2 uses a "dot" (.) notation rather than normal perl syntax, then tries to figure out what to do based on the type (I think it just looks at ref($whatever), but I am not very familiar its inner workings) when it compiles the template.
When you say something like [% d.e.f %], TT2 will actually use $d->e->f or $d->e->{f} or $d->{e}->{f}, etc., depending on what $d and $d->e (or $d->{e}, etc.) really is. This is the DWIM behavior that Randall referred to above.
| [reply]
[d/l]
[select] |